Abstract: A new method for nursery-leaf area measurement based on imagery processing is introduced.Using RGB tricolor gray value separation theory and 2R-G-B color character parameters to separate the background image，through the iteration theory calculating the best division threshold value，then leaf area is obtained by the leaf’s pixels distribution of image.Respectively measure the 72 leaves area in nursery by the traditional pane method and this method and confirm this method.The relevant analysis result shows that the correlation coefficient R²=0.992 1，the p value is 0.1E-4 when the significant level of 0.05，reaching an extremely remarkable level.This method has the high determination precision and satisfies the leaf area determination request.

Key words: machine vision, image processing, leaf area

摘要： 提出一种基于图像处理的穴盘苗叶面积测量方法。应用视觉图像系统采集穴盘图像，根据RGB三原色灰度值分离理论，采用2G－R－B颜色特征参数进行分割背景，通过迭代法求出图像最佳分割阈值，根据叶片像素点的分布比例计算叶片面积。采集穴盘中72个番茄叶片分别采用该文方法以及方格法测定叶片面积，对该文方法进行进一步验证，相关性分析结果表明，二者相关系数R²＝0.992 1，在0.05的显著水平时p值为0.1E-4，达极显著水平。文中方法具有较高测定精度，满足叶面积测定要求。

关键词: 机器视觉, 图像处理, 叶片面积